sapiopycommons 2025.8.12a693__py3-none-any.whl → 2025.8.12a696__py3-none-any.whl

sapiopycommons/callbacks/callback_util.py

@@ -1815,8 +1815,7 @@ class CallbackUtil:
         return response
 
     def request_file(self, title: str, exts: Iterable[str] | None = None,
-                     show_image_editor: bool = False, show_camera_button: bool = False,
-                     *, enforce_file_extensions: bool = True) -> tuple[str, bytes]:
+                     show_image_editor: bool = False, show_camera_button: bool = False) -> tuple[str, bytes]:
         """
         Request a single file from the user.
 
@@ -1826,8 +1825,6 @@ class CallbackUtil:
         :param show_image_editor: Whether the user will see an image editor when image is uploaded in this file prompt.
         :param show_camera_button: Whether the user will be able to use camera to take a picture as an upload request,
             rather than selecting an existing file.
-        :param enforce_file_extensions: If true, then the file extensions provided in the exts parameter will be
-            enforced. If false, then the user may upload any file type.
         :return: The file name and bytes of the uploaded file.
         """
         # If no extensions were provided, use an empty list for the extensions instead.
@@ -1847,12 +1844,11 @@ class CallbackUtil:
             file_path: str = self.__send_dialog(request, self.callback.show_file_dialog, data_sink=do_consume)
 
         # Verify that each of the file given matches the expected extension(s).
-        self.__verify_file(file_path, sink.data, exts if enforce_file_extensions else None)
+        self.__verify_file(file_path, sink.data, exts)
         return file_path, sink.data
 
     def request_files(self, title: str, exts: Iterable[str] | None = None,
-                      show_image_editor: bool = False, show_camera_button: bool = False,
-                      *, enforce_file_extensions: bool = True) -> dict[str, bytes]:
+                      show_image_editor: bool = False, show_camera_button: bool = False) -> dict[str, bytes]:
         """
         Request multiple files from the user.
 
@@ -1862,8 +1858,6 @@ class CallbackUtil:
         :param show_image_editor: Whether the user will see an image editor when image is uploaded in this file prompt.
         :param show_camera_button: Whether the user will be able to use camera to take a picture as an upload request,
             rather than selecting an existing file.
-        :param enforce_file_extensions: If true, then the file extensions provided in the exts parameter will be
-            enforced. If false, then the user may upload any file type.
         :return: A dictionary of file name to file bytes for each file the user uploaded.
         """
         # If no extensions were provided, use an empty list for the extensions instead.
@@ -1879,7 +1873,7 @@ class CallbackUtil:
         for file_path in file_paths:
             sink = InMemoryRecordDataSink(self.user)
             sink.consume_client_callback_file_path_data(file_path)
-            self.__verify_file(file_path, sink.data, exts if enforce_file_extensions else None)
+            self.__verify_file(file_path, sink.data, exts)
             ret_dict.update({file_path: sink.data})
 
         return ret_dict
@@ -1896,17 +1890,16 @@ class CallbackUtil:
         """
         if file_path is None or len(file_path) == 0 or file_bytes is None or len(file_bytes) == 0:
             raise SapioUserErrorException("Empty file provided or file unable to be read.")
-        if not allowed_extensions:
-            return
-        matches: bool = False
-        for ext in allowed_extensions:
-            # FR-47690: Changed to a case-insensitive match.
-            if file_path.casefold().endswith("." + ext.lstrip(".").casefold()):
-                matches = True
-                break
-        if not matches:
-            raise SapioUserErrorException("Unsupported file type. Expecting the following extension(s): "
-                                          + (",".join(allowed_extensions)))
+        if allowed_extensions:
+            matches: bool = False
+            for ext in allowed_extensions:
+                # FR-47690: Changed to a case-insensitive match.
+                if file_path.casefold().endswith("." + ext.lstrip(".").casefold()):
+                    matches = True
+                    break
+            if matches is False:
+                raise SapioUserErrorException("Unsupported file type. Expecting the following extension(s): "
+                                              + (",".join(allowed_extensions)))
 
     def write_file(self, file_name: str, file_data: str | bytes) -> None:
         """

sapiopycommons/flowcyto/flow_cyto.py

@@ -4,16 +4,38 @@ from weakref import WeakValueDictionary
 
 from databind.json import dumps
 from sapiopylib.rest.User import SapioUser
-from sapiopylib.rest.utils.singletons import SapioContextManager
 
 from sapiopycommons.flowcyto.flowcyto_data import FlowJoWorkspaceInputJson, UploadFCSInputJson, \
     ComputeFlowStatisticsInputJson
 
 
-class FlowCytoManager(SapioContextManager):
+class FlowCytoManager:
     """
     This manager includes flow cytometry analysis tools that would require FlowCyto license to use.
     """
+    _user: SapioUser
+
+    __instances: WeakValueDictionary[SapioUser, FlowCytoManager] = WeakValueDictionary()
+    __initialized: bool
+
+    def __new__(cls, user: SapioUser):
+        """
+        Observes singleton pattern per record model manager object.
+
+        :param user: The user that will make the webservice request to the application.
+        """
+        obj = cls.__instances.get(user)
+        if not obj:
+            obj = object.__new__(cls)
+            obj.__initialized = False
+            cls.__instances[user] = obj
+        return obj
+
+    def __init__(self, user: SapioUser):
+        if self.__initialized:
+            return
+        self._user = user
+        self.__initialized = True
 
     def create_flowjo_workspace(self, workspace_input: FlowJoWorkspaceInputJson) -> int:
         """

sapiopycommons/general/accession_service.py

@@ -5,7 +5,6 @@ from typing import Any
 from weakref import WeakValueDictionary
 
 from sapiopylib.rest.User import SapioUser
-from sapiopylib.rest.utils.singletons import SapioContextManager
 
 _STR_JAVA_TYPE = "java.lang.String"
 _INT_JAVA_TYPE = "java.lang.Integer"
@@ -275,10 +274,37 @@ class AccessionServiceDescriptor:
         }
 
 
-class AccessionService(SapioContextManager):
+class AccessionService:
     """
     Provides Sapio Foundations Accession Service functionalities.
     """
+    _user: SapioUser
+
+    __instances: WeakValueDictionary[SapioUser, AccessionService] = WeakValueDictionary()
+    __initialized: bool
+
+    @property
+    def user(self) -> SapioUser:
+        return self._user
+
+    def __new__(cls, user: SapioUser):
+        """
+        Observes singleton pattern per record model manager object.
+
+        :param user: The user that will make the webservice request to the application.
+        """
+        obj = cls.__instances.get(user)
+        if not obj:
+            obj = object.__new__(cls)
+            obj.__initialized = False
+            cls.__instances[user] = obj
+        return obj
+
+    def __init__(self, user: SapioUser):
+        if self.__initialized:
+            return
+        self._user = user
+        self.__initialized = True
 
     def accession_with_config(self, data_type_name: str, data_field_name: str, num_ids: int) -> list[str]:
         """

sapiopycommons/multimodal/multimodal.py

@@ -7,13 +7,35 @@ from weakref import WeakValueDictionary
 from databind.json import dumps, loads
 from sapiopylib.rest.User import SapioUser
 from sapiopylib.rest.pojo.DataRecord import DataRecord
-from sapiopylib.rest.utils.singletons import SapioContextManager
 
 from sapiopycommons.general.exceptions import SapioException
 from sapiopycommons.multimodal.multimodal_data import *
 
 
-class MultiModalManager(SapioContextManager):
+class MultiModalManager:
+    _user: SapioUser
+
+    __instances: WeakValueDictionary[SapioUser, MultiModalManager] = WeakValueDictionary()
+    __initialized: bool
+
+    def __new__(cls, user: SapioUser):
+        """
+        Observes singleton pattern per record model manager object.
+
+        :param user: The user that will make the webservice request to the application.
+        """
+        obj = cls.__instances.get(user)
+        if not obj:
+            obj = object.__new__(cls)
+            obj.__initialized = False
+            cls.__instances[user] = obj
+        return obj
+
+    def __init__(self, user:SapioUser):
+        if self.__initialized:
+            return
+        self._user = user
+        self.__initialized = True
 
     def load_image_data(self, request: ImageDataRequestPojo) -> list[str]:
         """

{sapiopycommons-2025.8.12a693.dist-info → sapiopycommons-2025.8.12a696.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: sapiopycommons
-Version: 2025.8.12a693
+Version: 2025.8.12a696
 Summary: Official Sapio Python API Utilities Package
 Project-URL: Homepage, https://github.com/sapiosciences
 Author-email: Jonathan Steck <jsteck@sapiosciences.com>, Yechen Qiao <yqiao@sapiosciences.com>
@@ -17,7 +17,7 @@ Classifier: Topic :: Scientific/Engineering :: Bio-Informatics
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Requires-Python: >=3.10
 Requires-Dist: databind>=4.5
-Requires-Dist: sapiopylib>=2025.7.31a279
+Requires-Dist: sapiopylib>=2025.4.17.264
 Description-Content-Type: text/markdown
 
 

{sapiopycommons-2025.8.12a693.dist-info → sapiopycommons-2025.8.12a696.dist-info}/RECORD

@@ -1,6 +1,6 @@
 sapiopycommons/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sapiopycommons/callbacks/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-sapiopycommons/callbacks/callback_util.py,sha256=Z1LcXnRRjXyhmcSDUwh4NzcA6ICtcbFUMKcvAqQcS8E,153811
+sapiopycommons/callbacks/callback_util.py,sha256=u-fXMp1tbIN8RHE4f5yG4pRVgz164XrqrGd-4ju2gYQ,153260
 sapiopycommons/callbacks/field_builder.py,sha256=rnIP-RJafk3mZlAx1eJ8a0eSW9Ps_L6_WadCmusnENw,38772
 sapiopycommons/chem/IndigoMolecules.py,sha256=7ucCaRMLu1zfH2uPIvXwRTSdpNcS03O1P9p_O-5B4xQ,5110
 sapiopycommons/chem/Molecules.py,sha256=mVqPn32MPMjF0iZas-5MFkS-upIdoW5OB72KKZmJRJA,12523
@@ -23,19 +23,17 @@ sapiopycommons/eln/experiment_tags.py,sha256=7-fpOiSqrjbXmWIJhEhaxMgLsVCPAtKqH8x
 sapiopycommons/eln/plate_designer.py,sha256=XFazSvhTbSy47t80-jc2tyx_-fQ_IUjKd18JQKEFcsY,13939
 sapiopycommons/eln/step_creation.py,sha256=CFkGC-SxwAQpQlcs_obqLAVgmsNxKSGMqMtO_E6IVmw,10171
 sapiopycommons/files/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-sapiopycommons/files/assay_plate_reader.py,sha256=3c2PQiiAbc2QJU9ZfNLzcTmvJrUwsbkIHO7R6R52xGU,3020
 sapiopycommons/files/complex_data_loader.py,sha256=T39veNhvYl6j_uZjIIJ8Mk5Aa7otR5RB-g8XlAdkksA,1421
 sapiopycommons/files/file_bridge.py,sha256=vKbqxPexi15epr_-_qLrEfYoxNxB031mXN92iVtOMqE,9511
 sapiopycommons/files/file_bridge_handler.py,sha256=SEYDIQhSCmjI6qyLdDJE8JVKSd0WYvF7JvAq_Ahp9Do,25503
 sapiopycommons/files/file_data_handler.py,sha256=f96MlkMuQhUCi4oLnzJK5AiuElCp5jLI8_sJkZVwpws,36779
-sapiopycommons/files/file_text_converter.py,sha256=Gaj_divTiKXWd6flDOgrxNXpcn9fDWqxX6LUG0joePk,7516
 sapiopycommons/files/file_util.py,sha256=djouyGjsYgWzjz2OBRnSeMDgj6NrsJUm1a2J93J8Wco,31915
 sapiopycommons/files/file_validator.py,sha256=ryg22-93csmRO_Pv0ZpWphNkB74xWZnHyJ23K56qLj0,28761
 sapiopycommons/files/file_writer.py,sha256=hACVl0duCjP28gJ1NPljkjagNCLod0ygUlPbvUmRDNM,17605
-sapiopycommons/flowcyto/flow_cyto.py,sha256=B6DFquLi-gcWfJWyP4vYfwTXXJKl6O9W5-k8FzkM0Oo,2610
+sapiopycommons/flowcyto/flow_cyto.py,sha256=vs9WhXXKz3urpjL8QKSk56B-NSmQR3O3x_WFBKoeO10,3227
 sapiopycommons/flowcyto/flowcyto_data.py,sha256=mYKFuLbtpJ-EsQxLGtu4tNHVlygTxKixgJxJqD68F58,2596
 sapiopycommons/general/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-sapiopycommons/general/accession_service.py,sha256=ZvtvZg7d_siMJUedjrF14mcqo5ZqVA5IJxDa5enlB-8,12792
+sapiopycommons/general/accession_service.py,sha256=3e__bVs7CYZ1CduLlGA9plnK7nCtdy7GXjCrNObPFgo,13484
 sapiopycommons/general/aliases.py,sha256=VwnWf_P803pcteoAIs0DkLScVChCS5XNgryTp8FzaNc,14696
 sapiopycommons/general/audit_log.py,sha256=sQAMcJx0cNkgZm7nTZSaGPxWvHG0_x6dBtU0jESavb4,9131
 sapiopycommons/general/custom_report_util.py,sha256=9elLEUSgfM0gli8nRPz1uYkhaXN4Vnx3piSiNHv5IBs,19156
@@ -47,7 +45,7 @@ sapiopycommons/general/popup_util.py,sha256=HKILegU1uCL_6abNlNL0Wn3xgX2JNa_kJeq7
 sapiopycommons/general/sapio_links.py,sha256=YkcVKNLrSGoM7tCCXBAsIbIxylctwdcEyhePrRMODe0,2859
 sapiopycommons/general/storage_util.py,sha256=ovmK_jN7v09BoX07XxwShpBUC5WYQOM7dbKV_VeLXJU,8892
 sapiopycommons/general/time_util.py,sha256=jU1urPoZRv6evNucR0-288EyT4PrsDpCr-H1-7BKq9A,12363
-sapiopycommons/multimodal/multimodal.py,sha256=EP9WYzx1CvidmEBlvzO6tiF4HJwsPB1FgxpnbWzxnpA,6161
+sapiopycommons/multimodal/multimodal.py,sha256=PFaGJPbKvW__tnxb8KkgkJZOKjQdgxF_kGfD5chet1s,6779
 sapiopycommons/multimodal/multimodal_data.py,sha256=0BeVPr9HaC0hNTF1v1phTIKGruvNnwerHsD994qJKBg,15099
 sapiopycommons/processtracking/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sapiopycommons/processtracking/custom_workflow_handler.py,sha256=eYKdYlwo8xx-6AkB_iPUBNV9yDoNvW2h_Sm3i8JpmRU,25844
@@ -64,7 +62,7 @@ sapiopycommons/webhook/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3
 sapiopycommons/webhook/webhook_context.py,sha256=D793uLsb1691SalaPnBUk3rOSxn_hYLhdvkaIxjNXss,1909
 sapiopycommons/webhook/webhook_handlers.py,sha256=7o_wXOruhT9auNh8OfhJAh4WhhiPKij67FMBSpGPICc,39939
 sapiopycommons/webhook/webservice_handlers.py,sha256=tyaYGG1-v_JJrJHZ6cy5mGCxX9z1foLw7pM4MDJlFxs,14297
-sapiopycommons-2025.8.12a693.dist-info/METADATA,sha256=QOtVYz4ZPmTHZQesBp8MtH3eGqepgB8wrzviPsS8jfw,3143
-sapiopycommons-2025.8.12a693.dist-info/WHEEL,sha256=C2FUgwZgiLbznR-k0b_5k3Ai_1aASOXDss3lzCUsUug,87
-sapiopycommons-2025.8.12a693.dist-info/licenses/LICENSE,sha256=HyVuytGSiAUQ6ErWBHTqt1iSGHhLmlC8fO7jTCuR8dU,16725
-sapiopycommons-2025.8.12a693.dist-info/RECORD,,
+sapiopycommons-2025.8.12a696.dist-info/METADATA,sha256=MXYSCdGuVmGhZYQwbpgQWlbvQtjy7YI6sTJMBD8poTo,3143
+sapiopycommons-2025.8.12a696.dist-info/WHEEL,sha256=C2FUgwZgiLbznR-k0b_5k3Ai_1aASOXDss3lzCUsUug,87
+sapiopycommons-2025.8.12a696.dist-info/licenses/LICENSE,sha256=HyVuytGSiAUQ6ErWBHTqt1iSGHhLmlC8fO7jTCuR8dU,16725
+sapiopycommons-2025.8.12a696.dist-info/RECORD,,

sapiopycommons/files/assay_plate_reader.py

@@ -1,93 +0,0 @@
-import base64
-import dataclasses
-from typing import Any
-
-from databind.core.dataclasses import dataclass
-from databind.json import loads
-from sapiopylib.rest.utils.singletons import SapioContextManager
-
-
-@dataclasses.dataclass
-class ProcessAssayPlateRequest:
-    """
-    A request to process the results of assay plate reader with a configuration set in Sapio.
-
-    Attributes:
-        num_rows (int): The number of rows in the plate.
-        num_columns (int): The number of columns in the plate.
-        plate_ids_in_context (list[str]): List of plate IDs that are in context for this request.
-        filename (str): The name of the file containing the assay data.
-        file_data (bytes): The binary content of the file.
-        plate_reader_config_name (str): The name of the plate reader configuration to use.
-    """
-    num_rows: int
-    num_columns: int
-    plate_ids_in_context: list[str] | None
-    filename: str
-    file_data: bytes
-    plate_reader_config_name: str
-
-    def to_json(self) -> dict[str, Any]:
-        return {
-            "numRows": self.num_rows,
-            "numCols": self.num_columns,
-            "plateIdsInContext": self.plate_ids_in_context,
-            "fileName": self.filename,
-            "fileDataBase64": base64.b64encode(self.file_data).decode('utf-8'),
-            "plateReaderName": self.plate_reader_config_name
-        }
-
-
-@dataclass
-class AssayPlateResultIdent:
-    plateId: str
-    channelIdOrBlock: str
-    kineticAssaySeconds: float | None
-
-
-@dataclass
-class AssayResultDatum:
-    """
-    Describes the data received from an assay plate reader.
-    Most of the time, the data is a single value, but sometimes it can be multiple values, especially for kinetic data.
-    """
-    DEFAULT_PROPERTY_NAME: str = "read"
-    rowPosition: str
-    columnPosition: str
-    valueByPropertyName: dict[str, float]
-    textValueByPropertyName: dict[str, str]
-
-
-@dataclass
-class AssayPlateResult:
-    """
-    Assay plate load result for a single plate in a file. A file can have more than one of this result if it has multiple plate of data in a single file.
-    """
-    resultIdent: AssayPlateResultIdent
-    numRows: int
-    numColumns: int
-    resultDatum: list[AssayResultDatum]
-
-
-@dataclass
-class AssayFileLoadResult:
-    """
-    The entire top-level file loading result for an assay plate reader file.
-    """
-    filename: str
-    plateResultList: list[AssayPlateResult]
-
-
-class AssayPlateReader(SapioContextManager):
-    """
-    This class contains services for Sapio Assay Plate Reader.
-    """
-
-    def process_plate_reader_data(self, request: ProcessAssayPlateRequest) -> AssayFileLoadResult:
-        """
-        Processes the assay plate reader data using provided request into a structured result using configuration defined in Sapio.
-        """
-        payload = request.to_json()
-        response = self.user.plugin_post("assayplatereader/process", payload=payload)
-        self.user.raise_for_status(response)
-        return loads(response.text, AssayFileLoadResult)

sapiopycommons/files/file_text_converter.py

@@ -1,207 +0,0 @@
-import io
-import os
-import tempfile
-from enum import Enum, auto
-
-class FileType(Enum):
-    """Supported file types for conversion."""
-    TXT = auto()
-    MD = auto()
-    CSV = auto()
-    DOC = auto()
-    DOCX = auto()
-    XLS = auto()
-    XLSX = auto()
-    PPT = auto()
-    PPTX = auto()
-    PDF = auto()
-    UNKNOWN = auto()
-
-
-class FileToTextConverter:
-    """
-    A class for converting various file types to raw text.
-    """
-    @staticmethod
-    def mime_type_to_enum(mime_type: str) -> FileType:
-        """
-        Converts a MIME type to a FileType enum.
-
-        :param mime_type: The MIME type string to convert.
-        :return: The corresponding FileType enum, or UNKNOWN if not recognized.
-        """
-        if not mime_type or not mime_type.strip():
-            return FileType.UNKNOWN
-
-        mime_map = {
-            "text/plain": FileType.TXT,
-            "text/markdown": FileType.MD,
-            "text/csv": FileType.CSV,
-            "application/msword": FileType.DOC,
-            "application/vnd.openxmlformats-officedocument.wordprocessingml.document": FileType.DOCX,
-            "application/vnd.ms-excel": FileType.XLS,
-            "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet": FileType.XLSX,
-            "application/vnd.ms-powerpoint": FileType.PPT,
-            "application/vnd.openxmlformats-officedocument.presentationml.presentation": FileType.PPTX,
-            "application/pdf": FileType.PDF,
-        }
-        return mime_map.get(mime_type, FileType.UNKNOWN)
-
-    @staticmethod
-    def file_extension_to_enum(file_path: str) -> FileType:
-        """
-        Converts a file path or extension to a FileType enum.
-
-        :param file_path: The file path or extension to convert.
-        :return: The corresponding FileType enum, or UNKNOWN if not recognized.
-        """
-        if not file_path or not file_path.strip():
-            return FileType.UNKNOWN
-
-        # Extract the file extension, removing the leading dot and making it lowercase
-        file_extension = os.path.splitext(file_path)[1].lstrip('.').lower()
-
-        ext_map = {
-            "txt": FileType.TXT,
-            "md": FileType.MD,
-            "csv": FileType.CSV,
-            "doc": FileType.DOC,
-            "docx": FileType.DOCX,
-            "xls": FileType.XLS,
-            "xlsx": FileType.XLSX,
-            "ppt": FileType.PPT,
-            "pptx": FileType.PPTX,
-            "pdf": FileType.PDF,
-        }
-        return ext_map.get(file_extension, FileType.UNKNOWN)
-
-    @classmethod
-    def parse_file(cls, file_type: FileType, file_bytes: bytes) -> str | None:
-        """
-        Parses file bytes based on the FileType and returns the text content.
-
-        :param file_type: The type of the file to parse.
-        :param file_bytes: The raw bytes of the file to parse.
-        :return: The text content of the file, or None if the file type is not supported or parsing fails.
-        """
-        if file_type is None or file_bytes is None:
-            return None
-        if not file_bytes:
-            return ""
-
-        # Dispatch to the correct parser method
-        parser_map = {
-            FileType.TXT: cls._parse_plain_text,
-            FileType.MD: cls._parse_plain_text,
-            FileType.CSV: cls._parse_plain_text,
-            FileType.DOC: cls._parse_doc,
-            FileType.DOCX: cls._parse_docx,
-            FileType.XLS: cls._parse_xls,
-            FileType.XLSX: cls._parse_xlsx,
-            FileType.PPT: cls._parse_ppt,
-            FileType.PPTX: cls._parse_pptx,
-            FileType.PDF: cls._parse_pdf,
-        }
-
-        parser_func = parser_map.get(file_type)
-
-        if parser_func:
-            return parser_func(file_bytes)
-
-        return None
-
-    @staticmethod
-    def _parse_plain_text(file_bytes: bytes) -> str:
-        return file_bytes.decode('utf-8')
-
-    @staticmethod
-    def _run_textract(file_bytes: bytes, extension: str) -> str:
-        """
-        Helper to run textract on in-memory bytes by writing to a temp file.
-        Note: textract may require external system dependencies.
-        """
-        import textract
-        with tempfile.NamedTemporaryFile(suffix=f".{extension}", delete=True) as temp_file:
-            temp_file.write(file_bytes)
-            temp_file.flush()  # Ensure all bytes are written to disk
-            text = textract.process(temp_file.name).decode('utf-8')
-        return text
-
-    @classmethod
-    def _parse_doc(cls, file_bytes: bytes) -> str:
-        return cls._run_textract(file_bytes, 'doc')
-
-    @staticmethod
-    def _parse_docx(file_bytes: bytes) -> str:
-        import docx
-        with io.BytesIO(file_bytes) as stream:
-            document = docx.Document(stream)
-            return "\n".join(para.text for para in document.paragraphs if para.text.strip())
-
-    @staticmethod
-    def _parse_xls(file_bytes: bytes) -> str:
-        import xlrd
-        workbook = xlrd.open_workbook(file_contents=file_bytes)
-        text_parts = []
-        for sheet in workbook.sheets():
-            text_parts.append(f"Sheet: {sheet.name}\n")
-            for row_idx in range(sheet.nrows):
-                row_cells = []
-                for col_idx in range(sheet.ncols):
-                    cell_text = str(sheet.cell_value(row_idx, col_idx))
-                    if cell_text.strip():
-                        row_cells.append(cell_text + "\t")
-                if row_cells:
-                    text_parts.append("".join(row_cells))
-                text_parts.append("\n")
-            text_parts.append("\n")
-        return "".join(text_parts)
-
-    @staticmethod
-    def _parse_xlsx(file_bytes: bytes) -> str:
-        import openpyxl
-        with io.BytesIO(file_bytes) as stream:
-            workbook = openpyxl.load_workbook(stream, read_only=True)
-            text_parts = []
-            for sheet in workbook.worksheets:
-                text_parts.append(f"Sheet: {sheet.title}\n")
-                for row in sheet.iter_rows():
-                    row_cells = []
-                    for cell in row:
-                        cell_text = str(cell.value) if cell.value is not None else ""
-                        if cell_text.strip():
-                            row_cells.append(cell_text + "\t")
-                    if row_cells:
-                        text_parts.append("".join(row_cells))
-                    text_parts.append("\n")
-                text_parts.append("\n")
-            return "".join(text_parts)
-
-    @classmethod
-    def _parse_ppt(cls, file_bytes: bytes) -> str:
-        return cls._run_textract(file_bytes, 'ppt')
-
-    @staticmethod
-    def _parse_pptx(file_bytes: bytes) -> str:
-        import pptx
-        with io.BytesIO(file_bytes) as stream:
-            presentation = pptx.Presentation(stream)
-            text_parts = []
-            for slide in presentation.slides:
-                for shape in slide.shapes:
-                    if shape.has_text_frame:
-                        text = shape.text_frame.text
-                        if text and text.strip():
-                            text_parts.append(text)
-            return "\n".join(text_parts)
-
-    @staticmethod
-    def _parse_pdf(file_bytes: bytes) -> str:
-        """Parses a PDF file's bytes and extracts text using PyMuPDF."""
-        import pymupdf
-        text_parts = []
-        with io.BytesIO(file_bytes) as stream:
-            with pymupdf.open(stream=stream) as doc:
-                for page in doc:
-                    text_parts.append(page.get_text())
-        return "\n".join(text_parts)